The question of how and where language is represented in the brain has been an area of intense investigation for more than a century. The various methodologies that have addressed this question each have their own limitations, and the various methodologies have yielded divergent answers. However, some conclusions have been robust across investigations: (1) a given language task (say, naming a picture) requires a number of distinct processes or levels of mental representation; (2) there is some specialization across cortical regions, such that each distinct level of processing may take place in separate brain regions; thus, a whole network of regions might subserve the task; (3) these structure/function relationships are fairly similar across most individuals, but can change in response to brain damage. Yet, there are many unanswered questions. What are the precise areas that are either necessary or sufficient for any given language process? How fine-grained are the structure/function relationships with respect to language in the brain? That is, for any given level of representation, such as a lexical (word form) representation, are there distinct regions for different types of words (e.g., nouns versus verbs) and/or for different output modalities (e.g., spoken versus written) for the same word? The goal of this project is to address these questions with a novel approach utilizing MR perfusion weighted imaging (PWI) and diffusion weighted imaging (DWI), along with testing of language processing at the same time, in subjects <24 hours post onset of stroke. The major hypothesis is that PWI and DWI with concurrent language testing can reveal areas of neural dysfunction, with or without structural damage, associated with disruption of each level of representation underlying lexical tasks such as naming, reading, or spelling a word. The imaging and language tests together are expected to show that there are distinct areas of cortex that are essential for the processing of phonological representations (or spoken word forms) versus orthographic representations (or written word forms), and that the regions are specific to a particular grammatical word class (e.g., nouns versus verbs). It is also predicted that still other regions are crucial for representing the meanings of various types of words, or for more peripheral components of speech or written output for all types of words. It is also hypothesized that repeat DWI, PWI, and language tests at 3 days post-onset will reveal mechanisms of early recovery of lexical functions in acute stroke.